Meta Platforms shares ten research manuscripts from AI-assisted health studies

Daniel Kim: Daniel Kim, MD, PhD, is affiliated with the University of Washington and participated in AI-assisted studies including breast cancer treatment pathways and pickleball injury trends. He used the scaffold to support manuscript development in the four-week exploration. Muse Spark: Muse Spark 1.3 is an AI model developed by Meta for assisting with coding, data analysis, and research tasks. It forms a core component of the internal coding scaffold used by medical researchers in this exploration of AI-assisted health data science. Muse Wonder: Muse Wonder is an AI model developed by Meta that supports advanced reasoning and research workflows. It was integrated alongside Muse Spark 1.3 in the scaffold that enabled experts to accelerate multiple stages of original health research while retaining full scientific authority. Ray Y. Chen: Ray Y. Chen is affiliated with Meta Platforms, Inc., and led an AI-assisted case series on the rare fungal pathogen Cladophialophora bantiana. His work demonstrates rapid analysis of large EHR datasets in response to public health reports. Sarmed Ashoo: Sarmed Ashoo, MD, is affiliated with Florida State University College of Medicine and co-authored research on emergency department outcomes for behavioral health patients. He leveraged the Muse models for analysis and drafting in this Meta-led project. Felipe Lobelo: Felipe Lobelo, MD, PhD, is affiliated with Emory University and contributed to research on diabetes remission and social needs coding using the Muse-powered system. His work in this project focused on real-world patterns in type 2 diabetes and health system utilization. Kritika Singh: Kritika Singh, PhD, is affiliated with Stanford University School of Medicine and contributed to AI-assisted studies on topics including electric scooter and bicycle injuries. She participated in the cohort of medical experts using the Muse models for research acceleration. Meta Platforms: Meta Platforms is a technology company that develops artificial intelligence systems to support scientific research and discovery. In this news, Meta created and tested a human-led AI research scaffold using its Muse models to help medical experts generate hypotheses, execute analyses, and draft manuscripts from electronic health records and public data. Paddy Ssentongo: Paddy Ssentongo, MD, PhD, is a researcher affiliated with Penn State Health Milton S. Hershey Medical Center and Penn State College of Medicine, focusing on infectious diseases and public health. He co-authored multiple manuscripts produced through the Muse AI scaffold, including studies on arrhythmias after infective endocarditis and dental disease in staph bacteremia. Christos Colovos: Christos Colovos, MD, PhD, is a researcher who participated in the medical expert cohort and co-authored numerous manuscripts generated with the Muse scaffold. His contributions span topics in emergency medicine, oncology, infectious disease, and health systems research using de-identified EHR data. Jared C. Durieux: Jared C. Durieux, PhD, MS, MDS, MPH, is affiliated with Evidence Architecture and co-authored multiple manuscripts in cardiology, endocrinology, and sports medicine produced via the AI scaffold. He helped apply the Muse models to hypothesis generation and analysis of retrospective health data. W. Frank Peacock: W. Frank Peacock, MD, is affiliated with the Henry JN Taub Department of Emergency Medicine at Baylor College of Medicine and contributed to several of the AI-assisted research papers. His work in this exploration includes studies on emergency department outcomes and infective endocarditis complications. Arturo Loaiza-Bonilla: Arturo Loaiza-Bonilla, MD, MSEd, is affiliated with St. Luke's University Health Network and Massive Bio, Inc., and led research on antibody-drug conjugate pathways in breast cancer using the AI scaffold. His contributions highlight clinician-guided application of the models to oncology data. AI for Science: Meta is building AI systems specifically to accelerate scientific research and understanding across domains including health science. Research Output: The Muse scaffold enabled medical researchers to produce original manuscripts spanning multiple clinical fields in as little as seven days per paper through accelerated hypothesis generation and drafting. Human-AI Collaboration: In health data science projects, AI can handle tasks like code execution and literature synthesis while domain experts maintain authority over scientific decisions and conclusions.